ENTROPY IN ECONOMICS: ITS MEASUREMENT THROUGH PRIMARY DATA AND ITS USEFULNESS IN BUSINESS

Transcription

1 ENTROPY IN ECONOMICS: ITS MEASUREMENT THROUGH PRIMARY DATA AND ITS USEFULNESS IN BUSINESS Elias Sanidas School of Economics, University of Wollongong Abstract Since Georgescu-Roegen s ( The entropy law and the economic process, 1971) pioneering work on entropy in economics, not many scholars have preoccupied themselves with this concept and its consequences. Recently some revival is taking place to link entropy and the business world (see for example The entropy vector, connecting science and business by Handscombe and Patterson, 24). Briefly entropy is the degree of disorder at some time for any system. For a closed system the natural tendency would be for entropy to keep increasing till final destruction. In the business world, entropy is also present and unless some actions are taken, any business will become more and more ineffective and eventually die. Sanidas ( The open system of four dynamic bio-socio-economic processes of the firm: the diamond of the black box, 26) has introduced four processes ( processes of the black box, or PROBB) that are negentropic, in other words they can prevent entropy from increasing during all operations of a firm. Based on Sanidas s work a questionnaire has been constructed and a survey was conducted with 12 firms in the marine industry in Australia. In this paper part of the results of this survey is presented with the aim to measure entropy in business. Thus, variables that denote waste directly or indirectly are some of the variables to correlate with all variables in the PROBB and inspect which such correlations are the strongest. The examination of these correlations together with some other analyses is providing us with some encouraging results as to the measurement of entropy. In turn such measurement will enable us to establish which areas are primary sources of inefficiencies. Obviously, the detection of these inefficiencies is necessary for firms to become more competitive and robust in their commercial and production activities. Key words: entropy, waste, measurement, firms Introduction Since Georgescu-Roegen s (1966, 1971) pioneering work on entropy in economics 1, several scholars have preoccupied themselves with this concept and its consequences. Recently some revival is taking place to link entropy and the business world; see for example Handscombe and Patterson s (24) paper as a general and simple exposition of the issue; Khalil (24) who made an overall critique of Georgescu-Roegen s work; Smith and Foley (28) who rigorously demonstrated the links between entropy and economic general equilibrium theory; Ayres (1994) who related entropy to many aspects of information, economics, evolution, and progress. Briefly, entropy is the natural universal process according to which there is a tendency for wasting energy so that increasing disorder and decreasing efficient productive work takes place in a system unless some actions occur that slow down this process. Handscombe and Patterson (24, p. 1) define entropy very simply as the degree of disorder or chaos that exists or is created. Thus, entropy in a closed system must remain constant or increase (Moore, 27, p. 38). 1 A good summary of this author s significant work can be found in Maneschi and Zamagni (1997). 44

2 From Handscombe and Patterson, (24), we can isolate some variables to represent entropy in business (see Table 1). Note that these variables although included in one of the three categories (columns in Table 1) can also be used in the other categories but in the opposite direction. Thus an entropy-increasing variable can also be an entropy-decreasing variable in a different context. A good example is leadership which can generate decisions that are either entropy-decreasing or the contrary. It would be useful to quote these authors for some of these variables. Business plans provide a way of trying to think ahead in order to organize (i.e. minimize) disorder in the way that a business is developed (ibid, p. 134). And so on. A company s ability to change is determined by its vision multiplied by its leadership and resources, all divided by the corporate age (ibid, p. 2); the corporate age is the inertia to do with culture, attitude, procedures, practices (ibid, p. 2). Information must be considered as a negative term in the entropy of a system We have only partial information and entropy measures the lack of information (ibid, p. 28). The entropy vector encourages us to realize that the lack of setting broad tolerances is as important as setting the specific objectives (ibid, p. 44. Experience curve, which is simply the continuous resetting of the entropy vector (ibid, p. 46). A lack of collaboration can be equated with entropy (ibid, p. 85). Table 1 Entropy-related factors/variables in conducting business Increasing entropy Decreasing entropy Increasing or decreasing entropy Waste Innovations Culture Inertia Experience Attitudes Lack of information Vision Procedures Decreasing constraints Leadership Risk taking thinking Deregulation Tolerance Lack of collaboration Objectives Stress level Production line Fatigue Knowledge Conformity/convention Links with outside Uncertainty Planning The aim of this paper is to check whether these factors (and others) as seen in Table 1 are important indeed in a real case. We will use as a real case a sample of 8 firms 2 from the marine industry in Australia (the survey was conducted during 26 and 27). For this purpose, we need a theoretical model which will help us to properly analyse the above real case in order to detect any signs of entropy. Recently Sanidas (25, 26) has introduced a comprehensive model of describing and explaining a firm s dynamic path in the economy. This model includes in a succinct way- all major factors that shape the firm s evolution in 2 By now 12 firms have been surveyed but the extra 4 on top of the 8 are not included in the analysis of this paper due to time limitations. 45

3 growth. The theory of the firm is extended to encompass all types of opportunity costs and not just transaction costs. As a further extension to the transaction costs and capabilities development and as a synthesis of several related issues, Sanidas (26) has introduced a complete system of 4 mutually exclusive, interdependent and negentropic 3 processes (PROBB) that fully describe the contents of the black box 4 of production (see Table 2). This model as Table 2 shows is a comprehensive summary of all elements that uniquely describe and explain the way firms are organized, managed and grow. These four PROBB are interdependent, although each PROBB contains unique elements that cannot belong to another PROBB. The important point to notice in this Table 2 with the PROBB variables is that each one of them represents opportunity costs. Thus, the POC variables represent transaction (or friction) costs, the POS elements stand for strategic opportunity costs, the POW variables express wisdom costs and the POM factors generate kinetic costs. All these different types of opportunity costs contribute to the natural law of entropy. However, it must be reminded that the PROBB variables can be both entropy-increasing and entropy-decelerating depending on how they are used in a business context. This will also be the object of investigation in the present paper. Table 2 The complete four processes of the black box (PROBB) P O W P O S P O M P O C Process of wisdom Process of strategies Process of movements Process of contracts Power Survival Infrastructure Superstructure Ability and memory Initiatives for action Movement relations Rules of the relations between the TIOP between the TIOP Wisdom costs* Strategic costs Kinetic costs Transaction costs Purpose*: to decrease Purpose: to produce Purpose: to produce less Purpose: to produce less negative * knowledge fewer mistakes waste* friction Potential energy Reaction energy Kinetic energy Friction energy Experience Strategies Timing Contracts with employees* Tacit knowledge Everyday decisions* Kinetic Procedures Legal form of the firm (e.g. in just-in-time) Education and training Planning Kinetic routines Contracts with suppliers* Culture and aesthetics Vision Layout Contracts with customers Information and data Mission Transport Contracts with society Competences and Objectives Teamwork Legal standards capabilities* R&D* Attacks Kinetic coordination Accounting rules (harmonization) Imitation Defense Implementation Institutions Innovations* Inertias Execution Governance 3 Negentropic means that the effect on entropy is negative, hence the underlying processes have the tendency to slow down increases in entropy or even to decrease entropy. 4 The term black box was coined by Coase (1992), the Nobel Prize laureate who introduced transaction costs in economics. 46

4 Leadership* Momentum Kinetic organization* Trust* POW POS POM POC Techniques of analysis Entrepreneurship* Effort non-physical Standards Needs* (e.g. for Domination and Effort physical* Authority exploration) Exploitation Motivation* Initiatives Fatigue Control Cognitive capacity and Inspiration for action Cooperation (actual Opportunism attention kinetic) Bounded rationality: Decision making* Work satisfaction as a team Supervision rules Subjectivity Idiosyncrasy Forecasting Kinetic tasks Incentives Attitudes and beliefs* Uncertainty Ergonomics Functions Marketing intelligence* Mistakes Logistics Autonomy Sophistication Policies Problem solving* Negotiations Socio-Psychological Organizational defensive Mechanisms of feedback* Documentation states* routines* Explicit knowledge Risk Performance Ownership rules Design of products Reactions to fortuitous events Ad hoc non-routine movements Informal rules POW POS POM POC Operations research Readiness and Operations research Conventions techniques emergence* applications (e.g. PERT) Intuition*, impressions, Interpretation and Work rationalization Conflict solutions perception* judgment (e.g. scientific Organizational spirit* Sense making* Location Communications rules and capital Imagination* and Improvisations and Projects Status Afflatus Heuristics* Unconscious* Expectations Transfer Hierarchy Illusions* Will Form or structure of the firm (e.g. M-form) Conception, insights Equity Subconscious and emotions* Utility versus virtue* *For the terms with an asterisk see Appendix 2 of Sanidas, 26. Source: Sanidas (26). The remaining of the paper will be as follows. Section 2 will introduce methodology and present the results; and section 3 will discuss conclusions. Methodology and results The model PROBB contains about 15 variables out of which about 12 are directly related to the elements of Table 2 above, and the remaining are more performance or economics related. The former will be called the X, Y, Z, and W PROBB variables 47

5 corresponding to the four processes respectively (POS, POW, POM, and POC); whereas the latter will be called the V variables. They are: V1: Customers dictate terms and requirements (the opposite of the 7-score Likert scale would be: we, the firm, dictate these terms) V2: Suppliers dictate terms and requirements (as for V1) V3: Rivals are weak (strong): V4: Power we, the firm, have in dictating prices and/or quantities V5: Low cost strategy (low, high) V6: Low price strategy (low, high) V7: Niche market strategy V8: Product quality V9: Product uniqueness V1: Availability of large market for our product V11: Adoption of technical innovations V12: Creation of technical innovations V14: Technology choice affected company operations (low, high) V15: Company operations affected technology choice V21: Sales growth in the last 5 years These V variables represent the consequences of the importance and direction of all PROBB variables. Thus we would expect that some PROBB variables-for a given firm- are rather negative in their action towards the Vs. Those PROBB variables that are not well related to Vs -for a given firm- can be termed the potentially entropy increasing and those PROBB variables that are well related to Vs can be termed the potentially entropy decreasing. First we can test the impact of each PROBB variable on the totality of the Vs. Second, later further below, we will test the impact of each PROBB variable on each individual V. Thus we first need a test that testifies the impact of a given PROBB variable on the totality of the V variables. Such a test is the Bartlett test of sphericity usually used in factor analysis. It shows how well all variables in a group are correlated with each other. It is calculated as a chi-square test. The way to use this test in our study is as follows: first we calculate the Bartlett chi-square for all V variables only. This value is Then we calculate the same test by adding to all V variables a PROBB variable. The increment in chi-square will obviously indicate how well the PROBB variable is correlated with all V variables together. According to the Bartlett test and the procedure just described, Table 3 shows the distribution of calculated chi-square for all PROBB variables. Table 3 Bartlett test for the impact of each PROBB variable on all Vs Value of chi-square Absolute frequency of PROBB elements Relative frequency Cumulative frequency (%) 188 (category A) (category B) (up to 193) (193 to 23) 235 (category C) (category D) (235 to 22) Table 4 shows the lowest (A and B categories) and highest (C and D categories) performance (as chi-square) of individual variables in being correlated with the basic performance variables Vs (considered in their totality). 48

6 Table 4 Lowest and highest chi-square (Bartlett test) for PROBB variables A B C D x x x x1 239 x y x x9 246 z y4 192 y17 23 x z z z y z z z y36 24 z z z33 24 y z w w1 231 z w w w2 212 w7 242 w w w3 227 w w w w average median Note: Categories A and B are the least related to the V variables hence they are the least-decreasing categories of entropy. Categories C and D are the most related to the Vs hence they are the most decreasing categories of entropy. For the A category we have the following PROBB variables as seen in Table 4. X1: inertia X23: interpreting surroundings Z19: achievement of production routines Z22: well defined tasks that involve movements Z29: fatigue Z34: location Z36: careful implementation of production operations W16: status (hierarchy) hindering management W32: management resolving conflict W33: equity (vital/crucial) An inspection of these PROBB variables shows that it makes sense that they are not well correlated with the group of V variables. Thus, for example, inertia, fatigue, status, and conflict resolving do not offer a substantial amount of information in promoting the Vs. Each one of these 1 PROBB variables is not capable in reducing or slowing down entropy and hence entropy will follow its natural tendency to increase. For the B category we have the following PROBB variables as seen in Table 4. X26: uncertainty Y21: culture in company influencing relationships in the firm Y4: accumulated knowledge or ad hoc processes enhancing innovations Z15: frequency of usage of OR (operations research) methods Z32: layout Z35: harmonization/coordination of production activities W15: demarcation of functions W22: structure of firm (e.g. departmental) assisting management W29: selfish staff more than expected by upper management W31: staff negotiations with management 49

7 Whatever we said about the category A PROBB variables we can also say about these B category variables but to a lesser degree. In addition, for both A and B categories we can see that many of their corresponding PROBB variables are approximations or representations for entropy factors as mentioned in Table 1 above: uncertainty, inertia, lack of collaboration, and so on. This can be seen more precisely in Table 5. Inertia Table 5 Correspondence between some entropy-related and PROBB variables (part 1) Lack of information Decreasing constraints Lack of collaboration Stress level Conformity Uncertainty procedures X1 X23 W33 W29 Z29 Y21 X26 Z19 Z32 Z15 W31 W15 Y4 Z22 Z34 W32 W16 Z35 W22 Z36 For the C category we have the following PROBB variables as already indicated in Table 4. X12: entrepreneurship versus cautious type of management X17: exploring large markets to secure a large market share Y17: marketing intelligence Z25: non-routine versus routine activities (valuable) Z28: monitoring physical efforts Z33: using ergonomics W1: detailed and rigorous contracts with employees W2: detailed and rigorous contracts with suppliers W3: detailed and rigorous contracts with customers For the D category we have the following PROBB variables as already indicated in Table 4. X1: using formal strategies X9: defence in the market X13: initiatives with far reaching consequences Y15: mental and thinking capacities Y 36: self-interest versus altruism of staff as a group Y44: replication by other firms of your knowledge of products and services Z16: using methods to rationalize work W7: traditions and institutions of staff members W13: company form (e.g. Pty) affecting functioning of the firm W25: using and applying control measures For both the C and D categories we can see that many of their PROBB variables are good proxies for the variables in Table 1 that are considered to reduce or slow down entropy. Thus, W1, W2, W3, W13, and W25 increase constraints on the system: this is exactly what a system needs in order to slow down entropy as was mentioned above regarding the variables of Table 1. In Table 6 we complete the correspondence between the PROBB variables of categories C and D and the variables affecting the slow down of entropy as per Table 1. 5

8 Increasing constraints Table 6 Correspondence between some entropy-related and PROBB variables (part 2) Experience Leadership Tolerance Stress level Production line Knowledge Links with outside W1 W7 X12 Y36 Z28 Z16 Y15 Y44 X1 W2 X13 X9 Z33 Z25 Y17 X17 W3 W13 W25 Planning Table 7 shows the distribution of the four PROBB in terms of the four categories of entropy impact as described in Table 2. Table 7 PROBB categories and entropy related intensity Entropy category POS POW POM POC Total A: most increasing B: increasing (less than in A) Total A and B: increasing (or least decreasing) C: decreasing D: most decreasing (more than in C) Total A and B: decreasing (or slowing down) Grand total (A+B+C+D) From Table 7 we can see that the POC is heavily represented in both the increasing and decreasing PROBB variables. The POM is only mostly present in the entropy-increasing categories. From the PROBB model viewpoint, this means that kinetic (POM) and friction (POC) energies are mainly responsible for affecting entropy in a substantial way. Now we turn to the testing of the impact of each PROBB variable on each V variable separately. Thus it remains to see which PROBB variables are most correlated with the performance variables Vs. Table 8 shows the results. Table 8 Significant correlations between PROBB variables and V variables V1 V2 V3 V4 V5 V6 V7 V8 V9 V1 V11 V12 V14 V15 V21 X1 * * * * * 5 X2 * * * 3 X3 * * * 3 X4 * * 2 X5 * * 2 X6 * * * 3 X7 * * * * * 5 X8 * * * * 4 total 51

9 X9 * * * * * * 6 X1 X11 * * 2 X12 * * * * * 5 X13 * * * * * * * * * 9 X14 * * * * * 5 X15 * * * * * * 6 X16 * * * * 4 X17 * * * * 4 X18 X19 * * * * * * 6 X2 * 1 X21 * * * * * * 6 X22 * 1 X23 * 1 X24 * * * 3 X25 * * * * 4 X26 * * 2 X27 * * * * * * 6 TOTAL POS V1 V2 V3 V4 V5 V6 V7 V8 V9 V1 V11 V12 V14 V15 V21 total Y7 * * * * * 5 Y8 * * * 3 Y13 * * 2 Y14 * 1 Y15 * * * * * * * * 8 Y17 * * * * 4 Y18 * * 2 Y19 * * * * * 5 Y2 * * * 3 Y21 * 1 Y22 * * * 3 Y23 * * 2 Y31 * 1 Y32 * * 2 Y33 * * * 3 Y34 * * * 3 Y35 * * 2 Y36 * * 2 Y38 * * * * 4 Y39 * * * 3 52

10 Y4 * * 2 Y41 * * 2 Y42 * 1 Y43 * * * 3 Y44 * * * * * * * * 8 Y5 * * * 3 Y51 * * * 3 TOTAL POW V1 V2 V3 V4 V5 V6 V7 V8 V9 V1 V11 V12 V14 V15 V21 total Z1 * * * 3 Z2 * * 2 Z15 Z16 * * * * 4 Z17 * * 2 Z18 * 1 Z19 * * 2 Z2 Z21 * * 2 Z22 Z23 * * 2 Z24 * * * * 4 Z25 * * * * * * 6 Z26 Z27 * 1 Z28 * * * * * 5 Z29 Z3 * 1 Z31 * * * * * * 6 Z32 Z33 * * * * 4 Z34 * 1 Z35 * * * 3 Z36 * * * 3 Z37 * * * 3 Z38 * * * 3 TOTAL POM V1 V2 V3 V4 V5 V6 V7 V8 V9 V1 V11 V12 V14 V15 V21 total W1 * * * * 4 W2 * * * 3 W3 * * * * * 5 W4 * * * 3 53

11 W5 * * * 3 W6 * * * * 4 W7 * * * * 4 W8 * * * * * * 6 W9 * * 2 W1 * * * * * 5 W11 * 1 W12 * * * * 4 W13 * * * * * 5 W14 * * * * 4 W15 W16 W17 * 1 W22 * * 2 W23 * * * * * 5 W24 * * * * * * * 7 W25 * * * * * * 6 W26 * * 2 W27 * * 2 W28 * * 2 W29 * * 2 W3 * 1 W31 * * * 3 W32 W33 * 1 TOTAL POC GRAND TOTAL ALL PROBB V1 V2 V3 V4 V5 V6 V7 V8 V9 V1 V11 V12 V14 V15 V Notes: (i) For the definition of the remaining PROBB variables see Appendix 1. The correlation coefficients are usually between.1 and.3. In Table 8 we can see that some variables V are more correlated with PROBB variables than other variables V. Thus, for example, V9 (product uniqueness) is significantly correlated with 35 PROBB variables. Or V1 (customers dictate terms and requirements) are significantly only correlated with 13 PROBB variables, and so on. These differences can be used to measure the degree of negentropy. Georgescu-Roegen (1966, p. 76) mentions Boltzmann s formula: entropy = k log n, where k is a constant and n is the number of equivalent microstates. Handscombe and Patterson (24, p. 38) use a simple example to demonstrate the meaning of Boltzmann s simplified formula: n can be for example 1 out of 5 (low entropy) possible patterns in an experiment or 1 out of 1 (high entropy). Consequently in our case, an approximation of direct calculation of entropy would be for V7: entropy = k log 33/19 = -.52k, or for V2: entropy = k log 17/19 = -.81k. Thus, entropy for V2 is greater than entropy for V7 (by considering the absolute value), and so on. total 54

12 In other words, the strategic/performance variable V2 is not sufficiently correlated with PROBB variables to guarantee low entropy as is the case with V7. This is the case for the sample of 8 firms surveyed in the marine industry in Australia, and another survey in a different sector might yield different results. As Georgescu-Roegen (1966, p. 78) sums it up for Boltzmann s formula 5 : If the higher the disorder the greater is the probability of its occurring, then obviously any closed system such as nature has a tendency to pass from any state to one of higher disorder. Table 9 summarizes the results of Table 8 in terms of the four PROBB and total. We can see -in this way- what the effect is of each PROBB on the V variables. Some of these effects will be discussed in the last section below. Thus, for example, the POS is the most influential entropy-wise on V11 (1 significant correlations), and so on. Table 9 Summary of the effects of the four PROBB on individual Vs V1 V2 V3 V4 V5 V6 V7 V8 V9 V1 V11 V12 V14 V15 V21 POS POW POM POC TOTAL These entropy-related computations for the Vs can be complemented with the frequency of each PROBB variable appearing in these computations. Thus, from Table 8 we can see that variables X13 appears 9 times as being significantly correlated with the Vs, Y15 appears 8 times, and so on. In this way we can determine once more which PROBB variables contribute to lowering entropy in the firms. Table 1 shows the most entropy-decreasing and the least entropy-decreasing PROBB variables. Table 1 The most entropy-decreasing and the least entropy-decreasing PROBB variables (in affecting the Vs individually) Most X13 Y15 Y44 W24 X9 X15 X19 X21 X27 Z25 Z31 W8 W25 Least X1 X18 Z15 Z2 Z22 Z26 Z29 Z32 W15 W16 W32 Y14 Y21 Note: For the most entropy-decreasing PROBB variables the frequency of significant correlations with Vs individually vary from 6 to 9 (mostly 6). For the least entropy-decreasing PROBB variables the frequency of significant correlations with Vs individually vary from to 1 (zero for first 11 out of 13 in the Table) We will finish this section by examining two special variables in the survey, those indicating waste of time (V*16) and negative emotions (V*2). Both these factors are a primary consequence of entropy increasing in any system and indeed in a firm and both indicate waste. The latter is the most obvious consequence of law of entropy (see for example the pioneering work by Georgescu-Roegen (1966, 1971). Table 11 shows which entropyrelated PROBB variables are strongly correlated with these two V*s. Table 11 High correlations between the waste and PROBB variables 5 In many economics applications a modern version of Boltzmann s formula is used: where S stands for entropy and f i stands for the fraction in microstate i (see Ayres, 1994, p. 36). 55

13 V*16 X5 X25 Y33 Y39 Y4 Z2 Z17 Z27 Z38 W3 W5 W7 W14 W29 W31 corr V*2 X5 X6 X1 X13 X24 X25 Y33 Y36 Z24 Z25 Z3 W2 W13 W24 W3 corr All signs of the correlation coefficients are the correct ones based on a priori considerations. For example, negative emotions (V*2) are negatively correlated with inertia (X1); this makes sense as a state of inertia (not enough actions) usually brings negative feelings, etc. These results will be further examined in the next section. Discussion of main results and conclusion Entropy is a well known phenomenon in physics. Social sciences have imitated its usefulness in the areas of economics, information theory, and so on, but without any substantial success. In the present study we consider entropy as a general natural tendency for waste and inefficiencies in business unless some measures are continually taken to slow down this tendency. This paper is only a modest attempt to quantify -in simple ways- the phenomenon of entropy in business. The latter takes various forms as seen in Table 1: waste, inertia, lack of information, and so on. In order to measure all these forms we used primary data of a survey of marine industry firms in Australia according to the PROBB (process of the black box ) model. This model is represented by 19 organizational cum managerial variables shown in Table 2 of the text. The PROBB variables can fully describe and explain the structure and functioning of a firm. The main task was to measure the extent to which these PROBB variables affect some strategic cum performance variables (the Vs in the text). In general, when the PROBB variables are positively and strongly correlated with the Vs then we suggest that they become entropy-decreasing 6 factors (and vice-versa). Thus, wasting time (V*16) in firms is positively (see Table 1) correlated (therefore reinforcing each other) with social relationships (Y33), imitating other firms (Y39), accumulating knowledge (Y4), measures to reduce work time (Z17), reliance on physical efforts (Z27), contracts with customers (W3), making rigorous contracts (W5), traditions and institutions (W7), and negotiations with management (W31). On the other hand, V*16 is negatively correlated (hence time-reducing) with use of formal objectives (X5), use of formal forecasts (X25), resolving problems as teams (Z2), performance according to customers (Z38), appreciation of governance (W14), and unselfish staff members (W29). It would be worth examining each one of these relationships in more depth but this is out of the scope of this paper. Nonetheless the interested readers can still do that at their own discretion and time. However, we must cross-check these results with the correlations between the above variables (in Table 11) and the strategic/performance variables Vs. Thus, making rigorous contracts (W5) is strongly and positively correlated (.28) with sales (V21). Consequently, if we calculate the partial correlation between sales and W5 by controlling for V*16 we get.3, which is higher as expected than.28 because we took away the negative influence of V*16. Similarly for other cases we can get the net (partial) correlations but in general the differences between direct and partial correlations are not large. For the negative emotions V*2 waste variable we have the following positive correlations with the PROBB ones: use of formal objectives 7 (X5), use of formal policies (X6), initiatives with 6 In other words these entropy-decreasing variables slow down the natural tendency for the firm to be disordered, or in every day language to be badly managed. 7 Note that the use of formal objectives (X5) enhances negative emotions (probably due to increased pressure to meet these objectives) whereas the same use of objectives decreases wasted time (negative correlation with V*16). 56

14 far reaching consequences (X13),, use of formal forecasts (X25), feedback mechanisms (Z24), contracts with suppliers (W2), firm form (W13), and international standards (W24). Thus, contracts with suppliers (W2) usually have the tendency to give headaches. On the other hand, negative emotions are enhanced by the following PROBB variables: inertia (X1), improvisations when deciding (X24), social relationships (Y33), self-interest oriented staff (Y36), non-routine operations (Z25), automatic cooperation (Z3), and performance incentives (W3). Thus, social relationships have the tendency to diminish negative emotions. From Table 1 we can pinpoint more precisely the PROBB variables that contribute the least (not significantly related to any of the Vs) to the entropy problem. Thus, inertia (X1), ad hoc decisions (X18), operations research methods (Z15), using transportation (Z2), tasks involving movements (Z22), transfers of people and equipment (Z26), fatigue problems (Z29), layout problems (Z32), functions (W15), status (W16), and resolving conflict (W32) do not assist at all in reducing entropy in a given firm. On the other hand those PROBB variables that contribute the most (significantly related to at least six of the Vs) to the entropy problem (in improving it) are: initiatives far reaching (X13), thinking capacity (Y15), imitation by other firms (Y44), international standards (W24), defending in markets (X9), rational decisions (X15), aggressive decisions re staff (X21), taking risks (X27), non-routine valuable operations (Z25), teamwork satisfaction (Z31), supervision rules documented (W8), and control measures (W25). Many of these variables, both most entropy decreasing and least entropy decreasing are also present in Table 4 of the text which shows the impact of PROBB variables on the totality of the V variables. Thus, X1, Z22, Z29, and others are the least entropy decreasing variables that affect the totality of the performance variables V. These common 8 PROBB variables in Tables 4 and 1 increase the confidence we have on the whole analysis. Other important conclusions can be drawn from Table 9 of the text which shows the importance of each PROBB on the V variables in terms of entropy. Thus, the POS (process of strategies, see Table 2) is very important in determining -hence entropy-decreasing- V4 (13 POS variables out of a total of 27 are significantly correlated with V4). This means that the way firms decide and form strategies (POS) have a significant impact on the power firms have in dictating prices and/or quantities (V4). POS has also a significant entropy effect on V11 (adoption of technical innovations) and sales (V21). The process of wisdom (POW) is very prominent in determining product uniqueness (V9); this makes sense as firms need wisdom and knowledge to create a unique product. The other two processes (POM and POC) do not seem to influence a particular V variable more prominently. Some general conclusions can also be drawn from the present study. First, it has become apparent that the PROBB model is consistent within itself, as all correlations examined in the context of entropy make sense. Second, although we have given more emphasis to the Vs in relation the PROBB variables, another way to determine the entropy impact on firms would have been to examine the impact of PROBB variables on themselves. This alternative is the object of another paper (where for example we could use structural equations models). Third, overall, we are confident that the present analysis has provided some evidence -in a simple way- about the impact of entropy-related variables on the performance and survival of firms. 8 It is expected that not all PROBB variables are present in both Tables 4 and 1 because the impact of these variables are either on the totality of Vs (hence Table 4) or on each V separately (Table 9). 57

15 Appendix 1: Definition of PROBB variables as per PROBB questionnaire POS (Process of POW (Process of POM (Process of POC (Process of Strategies) Wisdom) Movements) Contracts) X1: use of formal strategies Y7: using quant/e techniques Z1: using teamwork W1: contracts with employee X2: use of formal planning Y8: experience knowledge Z2: resolve problems as team W2: contracts with suppliers X3: use of formal vision Y13: education/ training Z15: using operations research W3: contracts with customers X4: use of formal mission Y14: information/ data Z16: rationalize work W4: support community X5: use of formal objectives Y15: thinking capacity Z17: reduce work time W5: rigorous contracts/documents X6: use of formal policies Y17: marketing intelligence Z18: link work stages W6: rigorous accounting rules X7: expectations reflecting markets Y18: transmitted knowledge Z19: production routines W7: traditions & institutions X8: aggression in markets Y19: product design Z2: using transportation W8: supervision rules documented X9: defending in markets Y2: insightful staff Z21: movements organ/on W9: ownership rules assistance X1: Inertia state Y21: culture in relationships Z22: movements tasks W1: informal v formal rules X11: momentum Y22: rational thinking Z23: logistics W11: conventional behaviour X12: cautious style management Y23: attitudes/beliefs of staff Z24: feedback mechanisms W12: rules of communication X13: initiatives far reaching Y31: socio-psycho/al values Z25: non-routine operations W13: firm form affecting you X14: reaction unexpected Y32: intuition/perceptions Z26: transfers people and equipment W14: governance appreciation X15: rational decisions Y33: social relationships Z27: physical efforts W15: functions reliance X16: driven outside Y34: imagination Z28: physical efforts W16: status hindering square constructive monitoring X17: decisions market Y35: emotional state Z29: fatigue problems W17: hierarchy structure share X18: ad hoc decisions extent Y36: self-interest of staff Z3: automatic cooperation W22: firm structure assisting X19: Readiness to adversity Y38: distinct capabilities Z31: teamwork satisfaction W23: standard legal documents X2: easy decisions extent Y39: imitating other firms Z32: layout problems W24: international standards X21: aggression re staff Y4: accumulated Z33: ergonomics W25: control measures knowledge X22: asses/t before decisions Y41: leadership vision Z34: location of activities W26: autonomy employees X23: interpreting surroundings Y42: leadership motivation Z35: harmonized activities W27: trust your employees X24: improvisations Y43: unique culture Z36: implementation W28: attitude re authority deciding prod/n X25: use of formal forecasts Y44: other firms imitating you Z37: execution as expected W29: selfish staff members X26: uncertainty facing Y5: needs in markets Z38: performance as per customs W3: performance incentives X27: taking risks Y51: outdated knowledge W31: negotiating with man/t W32: resolving conflict W33: equity if crucial Total : 27 Total: 27 Total: 26 Total: 29 58

16 References 1. Ayres, R.U. (1994), Information, Entropy, and Progress, AIP Press, New York 2. Coase, R.H. (1992), The institutional structure of production, American Economic Review, September, 82, Georgescu-Roegen, N. (1966), Analytical Economics, Harvard University Press, Cambridge 4. Georgescu-Roegen, N. (1971), The Entropy Law and the Economic Process, Harvard University Press, Cambridge 5. Handscombe, R.D. and Patterson, E.A. (24), The Entropy Vector, Connecting Science and Business, World Scientific, New Jersey 6. Khalil, E.L. (24), The Three Laws of Thermodynamics and the Theory of Production, Journal of Economic Issues, 38 (1), Maneschi, A. and Zamagni, S. (1997), Nicholas Georgescu-Roegen, , The Economic Journal, 17, May, Sanidas, E. (25), Organizational Innovations and Economic Growth, Organosis and growth of firms, sectors, and countries, Edward Elgar, Cheltenham 9. Sanidas, E. (26), The open system of four dynamic bio-socio-economic processes of the firm: the diamond of the black box, The Journal of Socio-Economics, 35(3), Smith, E. and Foley, D.K. (28), Classical Thermodynamics and economic general equilibrium theory, Journal of Economic Dynamics and Control, 32,